Traditionally, students experience difficulties solving problems involving geometric shapes and their properties. In line with the current curricular guidelines, it's important to reflect about the use of active learning strategies, which directly engage students in meaningful mathematical activity, that contribute to reverse this situation. This paper refers to a study that aims to understand the influence of peer written feedback on students' performance in a Gallery Walk context, in the scope of problem solving involving areas of plane figures. We followed a qualitative, interpretative approach and collected data in a 5th grade class, with 19 students, through participant observation, a questionnaire, interviews, written productions and photographic records. Results show that, when the sender of feedback is a student, different types of feedback are used, however not all of them promote the improvement of the receiver's performance. It was noticed that the most effective feedback is focused mainly on the emerging mathematical content, commenting on the student's performance in terms of task solution, process and results. It was also found that the feedback must be clear and simple so that the receiver clearly understands what the sender wants to convey.